LimeSDR decreasing gain over time

I have a LimeSDR USB-A for use as a compact radio telescope receiver, so I’m looking at the 1420 MHz region. There are two cascaded external preamps and I don’t use the onboard LNA. I observe that the gain diminishes so that signal is halved within 20 minutes of power on and continues to decrease at a decelerating rate. As that happens the ‘DC’ peak appears and increases in height. I note that the board warms over that time - reports 51C when it stabilizes. I have heat sinks on the receiver and FPGA chips. The system recovers to the initial state when I disconnect the LimeSDR from power (USB), and preamps are kept on.
Has this been seen before? Perhaps a component is failing and sensitive to heating? I do have heat sinks on the receiver and FPGA chips.

I keep my Lime SDRs on 40 C level and bellow… Try that pls.

Update: An immersion cooling system has worked for me where heat sinks would not.

What kind of coolant you used?
On my side I am using now one LimeSDR USB in passive cooling mode.
Small coolers on LMS7 and FPGA and 3 layers of 1 mm silicon thermal pads
bellow the board. Average reading after 1 h on RX is 35-36 C @ Room temp. 22 C

I use white food-grade mineral oil. Based on literature, one paper I’ve read that compared several fluids, it is sufficiently inert. Even then, cooling the container to around 15C is beneficial with respect to receiver gain. Lower and the clock oscillators appear to fail. This is the case with LimeSDR USB. With the USB Mini I can cool the container to 0C with gain improvement and no failures. As another benefit, I can routinely run the Mini at a sample rate of 60,000,000sps, bandwidth 20,000,000 with performance like the larger LimeSDR USB.

1 Like